Chapter 3 Pressure-activated membrane processes M. J. LEWIS, Department of Food Science and Technology, The University of Reading, Reading RG6 6AP 3.1 INTRODUCTION Over the last 30 years, a number of membrane processes have evolved, which make use of a pressure driving force and a semi-permeable membrane in order to effect a separa- tion of components in a solution or colloidal dispersion. The separation is based mainly on molecular size, but to a lesser extent on shape and charge. The three main processes are reverse osmosis (hyperfiltration), ultrafiltration and microfiltration. The dimensions of the components involved in these separations are given in Fig. 3.1, and are typically in the range of less than 1 nm to over 1000 nm. A brief summary of the main differences between them, in terms of the components which are rejected by the membranes, is also illustrated. More recently the process term ‘nanofiltration’ has been introduced, which is somewhere between reverse osmosis (RO) and ultrafiltration, bringing about a separation of low molecular weight components such as monovalent ions and salts from organic MF Solutions Macromolecules Fat globules --- Ions, sugars Proteins Suspensions - 1 I I I I I I 10-2 1 102 io4 (nm) Fig. 3 1 Size ranges for different membrane processes 66 M. J.Lewis compounds such as sugars. These pressure-activated processes can also be regarded as a continuous spectrum of processes, with no obvious distinct boundaries between them. However, it should be noted that the sizes of the components being separated range over several orders of magnitude, so it is highly likely that the separation mechanisms and hence the operating strategies may change as we move through the spectrum. 3.2 TERMINOLOGY The feed material is applied to one side of a membrane. The feed is usually a low- viscosity fluid, which may sometimes contain suspended matter and which is subjected to a pressure. In most cases the feed flows in a direction parallel to the membrane surface and the term cross-flow filtration is used to describe such applications. Dead-end systems are used, but mainly for laboratory scale separations. The stream which passes through the membrane under the influence of this pressure is termed the permeate (filtrate). After removal of the required amount of permeate, the remaining material is termed the con- centrate or retentate. The extent of the concentration is characterised by the concentration factor df), which is the ratio of the feed volume to the final concentrate volume (see equation (3.5)). The process can be illustrated simply in Fig. 3.2(a). From a single membrane process- ing stage, two fractions are produced, named the concentrate and permeate. The required extent of concentration may not be achieved in one stage, so the concentrate may be returned to the same module for further concentration or taken to other modules in a cascade, or multistage process. The permeate may also be further treated in a separate process. In terms of size considerations alone, one extreme is a membrane with very small pore diameters (tight pores). In this case the permeate will be pure water because even small molecular weight solutes will be rejected by the membrane; high-pressure driving forces are required to overcome frictional resistance and osmotic pressure gradients. If the permeate is predominantly water, then the process is known as reverse osmosis or hyperfiltration; it is similar in its effects to evaporation or freeze-concentration. A con- centrate will be produced, in which there is virtually no alteration in the proportion of the Concentrate Feed :-+ pi L 1: IR I Permeate ++- Osmosis + Reverse osmosis I (4 (b) Fig. 3.2. Separation of feed into a concentrate and permeate stream. Pressure-activated membrane processes 67 solid constituents. In some applications it is the permeate which is the required material; for example the production of ‘drinking water’ from sea-water or ‘pure water’ from brackish water. The best processes are those where both the concentrate and the permeate are fully utilised. There have been several comparisons made between evaporation and reverse osmosis, in terms of capital costs, energy costs and product quality (Renner, 1991). In general terms RO is less energy intensive and can improve product quality. Some limitations are the high capital costs, membrane replacement costs and extent of concentration, which is not as high as that obtainable by evaporation. If a fluid, for example milk, is separated from water by a semi-permeable membrane (see Fig. 3.2(b)), there will be a flow of water from the water to the milk, in order to equalise the chemical potential of the two fluids; this is termed osmosis. This flow of water can be stopped by applying a pressure to the milk. This pressure that stops the flow is termed the osmotic pressure. If a pressure greater than the osmotic pressure is applied, the water will flow from the milk to the water, thereby reversing the natural process of osmosis and achieving a concentration of the milk. Therefore in reverse osmosis, the pressure applied needs to be in excess of the osmotic pressure. Osmotic pressure (T) is a colligative property, the pressure being dependent upon the number of particles and their molecular weight. In classical terms it is determined from the Gibb’s free energy equation: (3.1) RT ;rl=-1nyX “m where R = gas constant, T = absolute temperature, y= activity coefficient, X = mole frac- tion, and V, = partial molar volume. For dilute solutions of non-ionisable materials, the Van’t Hoff equation can be used z = RT(C/M) (3.2) where c = concentration (kg m-3) and M = molecular weight. For ionisable salts this becomes ;rl = iRT(c/M) (3.3) where i = the degree of ionisation, e.g. for NaC1, i = 2; for FeC12, i = 3. This equation predicts a linear increase in osmotic pressure with concentration. How- ever, this relationship breaks down, even at relatively low concentrations, with the rela- tionship between osmotic pressure and concentration becoming non-linear. For example, the osmotic pressure of a 25% serum albumin solution was 300 2, which is about six times higher than predicted from the Van’t Hoff equation. It is also affected by pH. This non-linear relationship can be represented by Virial type equations: ;rl = Ac+ Bc2 + Dc3 (3.4) where c = concentration and A, B and D are constants. The constants are presented for dextran and whey by Cheryan (1986). 68 M. J.Lewis Osmotic pressures are highest for low molecular weight solutes, so the highest osmotic pressures arise for salt and sugar solutions. Concentration of such solutions results in a large increase in their osmotic pressure. On the other hand, proteins and other macromol- ecules do not produce high osmotic pressures. There will only be small increases during their concentration as well as small differences in osmotic pressure between the feed and permeate in ultrafiltration. Values for osmotic pressures are not easy to find in the literature and a selection of values is given in Table 3.1. A further complication with foods and other biological systems is their complexity, with not just one but many components. In reverse osmosis the applied pressure must exceed the osmotic pressure, and the driving-force term in reverse osmosis is normally the difference between the applied pressure and the osmotic pressure. It could be that osmotic pressure is one of the factors that limits the extent of concentration. One suggested experimental method for measuring osmotic pressure is to determine the pressure that would give zero flux, by extrapolation. In ultrafiltration and microfiltration, there is little osmotic pressure differ- ence over the membrane as the low molecular weight components are almost freely permeating (see equation (3.8)). Table 3.1. Osmotic pressures of some solutions So I LI ti on Osmotic pressure (bar) Sugar beet 20" Brix 34.1 Tomato paste 33" Brix 69.0 Apple juice 15" Brix 20.4 Citrus juice 10" Brix 14.8 34" Brix 69.0 Sucrose 44" Brix 69.0 Coffee extract 28% TS 34.0 Sea-water 3.5% salt 23.2 15.0% salt 138.0 Milk 6.9 Whey 6.9 Lactose 1% w/v 3.7 Compiled from data in Cheryan (1986) and Lewis (1982). Some equations for osmotic pressure are given by Cheryan (1986). As the membrane pore size increases, the membrane becomes permeable to low molecular weight solutes in the feed; even the transport mechanisms are likely to change. Lower pressure driving forces are required as osmotic pressure differences between the feed and permeate are reduced. However, molecules of a larger molecular weight are still rejected by the membrane. Therefore some separation of the solids present in the feed takes place; the permeate contains low molecular weight components at approximately the same concentration as they are in the feed, and the concentrate contains large Pressure-activated membrane processes 69 molecular weight components at an increased concentration, compared to the feed. Note that some of the low molecular weight components will be retained in the concentrate. It is this fractionation and concentration process that makes the ultrafiltration process more interesting than reverse osmosis, although, as mentioned earlier, there is no sharp demarcation between the processes. More porous membranes still allow not only sugars and salts, but also macromolecules, to pass through, but retain particular matter and fat globules, i.e. greater than 100 nm (see Fig. 3.1); this is termed microfiltration. Because of their increased potential for separating components in mixed feeds, ultrafiltration and microfiltration are covered in more detail in Chapters 4 and 5. However, much of the discussion, particularly that on membrane performance and rejection, will also be pertinent to all three pressure-activated processes. Major points of difference are discussed later in this chapter. 3.3 CONCENTRATION FACTOR AND REJECTION Two important processing parameters for all pressure activated processes are the conceir- tration factor cf) and the membrane rejection characteristics. The concentration factor is defined as follows: (3.5) Concentration factor (f) = VF 1 V, where VF = feed volume and V, = final concentrate volume. The term volume reduction factor (VRF) is sometimes used: VRF = 1oo(VF - V,)/~F = 100(1- 1/f) (3.6) Thus a process with a concentration factor of 10 would have a volume reduction factor of 90%. The permeate volume (V,) equals the feed volume minus the concentrate volume (assuming no losses) (3.7) As soon as the concentration factor exceeds 1, the volume of permeate will exceed that of the concentrate. Concentration factors may range from as low as 1.5 for some viscous materials, to up to 50 for dilute protein solutions, e.g. chhana whey (Jindal and Grandison, 1992). Generally higher concentration factors are used for ultrafiltration than for reverse osmosis, e.g. up to 25-30 for UF of cheese-whey, compared to 5 for RO of cheese-whey. A mass balance for the process can be applied and is useful for estimating the distribution of components between the permeate and concentrate, or for estimating the losses that are incurred in practical situations. vp = v, - v, The rejection or retention factor (R) of any component is defined as R = CF - cp/cF (3.8) where CF is the concentration of component in the feed and cp is the concentration in the permeate. 70 M. J.Lewis It can be determined experimentally for each and every component in the feed, by sampling the feed and permeate at the same time and analysing the component in ques- tion. It is a very important property of a membrane, as it will influence the extent (quality) of the separation that can be achieved. Rejection values normally range between 0 and 1; sometimes they are expressed as percentages (0-1 00%). when cp = 0; when cp = CF R = 1; all the component is retained in the feed R = 0; the component is freely permeating. An ideal RO membrane would give a rejection value for all components of 1, whilst an ideal UF membrane, being used to concentrate a high molecular weight component or remove a low molecular weight component would give respective rejection values of 1 and 0. If the concentration factor and rejection value are known, the yield of any component, which is defined as the fraction of that component present in the feed, which is recovered in the concentrate, can be estimated. Obviously for reverse osmosis, the yield for an ideal membrane is 1.0. Rejection data for membranes and their effects on yield and separation performance will be discussed in greater detail in Chapter 4. 3.4 MEMBRANE CHARACTERISTICS The membrane itself is crucial to the process. The first commercial membranes were made of cellulose acetate and these are termed first-generation membranes. For food- processing applications, they had some limitations, with temperatures below 30°C and a pH range of 3-6. These were followed in the mid-1970s by other polymeric membranes (second-generation membranes), with polyamides and, in particular, polysulphones being widely used for foods. The resulting improvements in cleaning and hygiene are covered in Section 3.8. It is estimated that over 150 organic polymers have now been investigated for membrane applications. Inorganic membranes based on sintered and ceramic materials are also now available. The physical structure of these membranes is complex, and as most of them are used for microfiltration, their structure is described in more detail in Chapter 5. The main terms used to describe membranes are microporous or asymmetric. Microporous membranes have a uniform porous structure throughout, although the pore size may not be uniform across the thickness of the membrane. They are usually charac- terised by a nominal pore size and no particle larger than this will pass through the membrane. In contrast to this, most membranes used for ultrafiltration are of asymmetric type, having a dense active layer or skin of 0.5-1 pm in thickness, and a further support layer which is much more porous and of greater thickness (Fig. 3.3). Overall the porosity of these membranes is high, although the surface porosity may be low, with quoted values in the range 0.3-15% (Fane and Fell, 1987). Often the porous path may be quite tortuous, the distance covered by the solvent or solute being much greater than the thickness of the membrane; the term tortuosity has been used as a measure of this property. The pores are not of a uniform size, as can be seen when viewed under the electron microscope, and are best characterised by a pore size distribution. This 72 M.J.Lewis temperature and inducing more turbulence increases the flux. However, the flux is only affected by the applied pressure in the pressure-dependent region. These factors are discussed in more detail in Chapter 4. The power utilisation (W) is related to the pressure (head) developed and the mass flow rate as follows: W = m'hg (3.9) where m' = mass flow rate (kg s-'), h = head developed (nz) and g = acceleration due to gravity (9.81 m s-*). This energy is largely dissipated within the fluid as heat and will result in a tempera- ture rise. Cooling may be necessary if a constant processing temperature is required. The membrane offers a resistance to the transfer of both solvent (normally water) and solute. The permeate flux is a measure of the flow rate of solvent through the membrane, whereas the rejection describes the amount of solute which passes through (see eq. (3.8)). From a process engineering standpoint, it is highly desirable to be able to predict the flux and rejection from the physical properties of the solution, the membrane characteristics and the hydrodynamics of the flow situation, in order to optimise the performance of the system. Membrane operations have been subject to a number of modelling processes, in order to achieve these objectives. However, before these models are discussed in more detail, it is important to consider the phenomena of concentration polarisation and fouling. 3.6 TRANSPORT PHENOMENA AND CONCENTRATION POLARISATION A very important consideration for pressure-driven membrane processes is that the sepa- ration takes place not in the bulk of solution, but in a very small region close to the membrane, known as the boundary layer, as well as over the membrane itself. This gives rise to the phenomenon of concentration polarisation over the boundary layer. (Note that in streamline flow the whole of the fluid will behave as a boundary layer.) It is mani- fested by a quick and significant reduction (2-10 fold) in flux when water is replaced by the feed solution, for example in a dynamic start. Concentration polarisation occurs whenever a component is rejected by the membrane. As a result, there is an increase in the concentration of that component at the membrane surface, together with a concentration gradient over the boundary layer. Eventually a dynamic equilibrium is established, where the convective flow of the component to the membrane surface equals the flow of material away from the surface, either in the permeate or back into the bulk of the solution by diffusion, due to the concentration gradient established. This increase in concentration, especially of large molecular weight components, offers a very significant additional resistance. It may also give rise to the formation of a gelled or fouling layer on the surface of the membrane (see Fig. 3.4). Whether this occurs will depend upon the initial concentration of the component and the physical properties of the solution; it could be very important as it may affect the subsequent separation performance. Concentration polarisation itself is a reversible phenomenon; thus if the solution is then replaced by water, the original water flux should be restored. However, this rarely occurs in practice due to the occurrence of fouling, II II II 11 I1 I1 74 M.J.Lewis Thus the diffusivity and membrane thickness and its partition coefficient all influence the transport of solute. (Convective flow of solute is ignored at high solute rejections.) According to this model, increasing the pressure will have a preferential effect on solvent flow, thereby reducing solute concentration in the permeate and increasing rejection. It also predicts that increasing the solute concentration in the feed preferentially increases solute transport and increases the concentration in the permeate, thereby decreasing rejection. Increasing the temperature increases both solvent flux and solute flux by about the same amounts, thereby leaving the rejection unchanged. A 1°C change in temperature changes the flux (solute and solvent) by approximately 3%. Although many of these trends are observed with simple solutions, the theory does not account for all the observed facts for multi-component solutions, whereby the presence of one component increases the permeability of other components. Glover (1985) pointed out that this theory does not describe the behaviour of complex systems precisely, but it gives a background of understanding. In multicomponent systems, differences in the partition coefficients could explain the difference in permeabilities between different components. A second proposed model is the preferential adsorption, capillary flow model, which predicts that the component concentrated preferentially in the permeate will be that component which is adsorbed most strongly on the membrane surface. The component that is preferentially adsorbed onto the membrane surface provides a thin layer of that component adjacent to the surface. This thin layer then moves through the pores of the membrane by capillary flow, under the influence of a pressure gradient, and in this way is thus preferentially transported through the membrane. For membranes of a hydrophilic nature, the component preferentially absorbed and transported is water. Thus it is postulated that there is a thin film of water adjacent to the membrane surface. This theory also explains the low rejections and sometimes negative rejections for highly polar organic solutes, found with cellulose acetate membranes, due to their preferential adsorption on the surface. Other situations, where rejection of organic solutes decreases with increasing pressure, are explained by their adsorption onto the more hydrophobic regions of the membrane. A third model is based upon the wetted surface mechanism, whereby water adsorbs onto the surface of the membrane by hydrogen bonding. It is postulated that these clusters of water prevent solute entering the pores and that the water passes through the mem- brane from one adsorbed site to the next. The energy requirements for water migration are much less than salt migration, thereby promoting separation of the salt and water. All these models are based on knowing the transport mechanisms involved. The physical chemistry of a wide variety of membrane materials, including permeability data, diffusion data and sorption data which are required for the models described earlier, have been reviewed by Tsujita (1992). Also reviewed are transport properties related to non- equilibrium thermodynamics with uncharged and charged membranes. Other models are based upon irreversible thermodynamics, where the driving force for transport of solvent and solute is expressed in terms of differences in their chemical potential over the membrane. However, the fluxes for solvent and solute are coupled and the flux for each component is influenced by the chemical potential difference for both components. With such models, the exact mechanisms are not known, but the Pressure-activated membrane processes 75 relationship between the fluxes and driving forces are described by phenomenological coefficients, which are nevertheless connected with the mechanism. Such phenomenological models provide an empirical description of the transport of molecules through membranes, but they do not give an explanation for the separation mechanism. Gekas (1992) discusses these types of model in more detail for membrane processes. Cheryan (1989) has concluded that none of the proposed models can satisfactorily explain the rejection characteristics of all solvent-solute-membrane reverse osmosis sys- tems. On the other hand, ultrafiltration is regarded as a sieving process and is often mod- elled as a bundle of capillaries, with flow through the capillaries being described by the Hagen-Poiseuille equation: Q = d4 AP n/128pL (3.10) where Q = volumetric flow rate, AP = pressure drop; p = dynamic viscosity, L = capillary length and d = capillary diameter. (Note that this equation is often expressed as flux rate/unit area (d2AP/32pL).) Additional terms may be introduced to account for porosity and tortuosity effects in the membrane. These equations predict that the flow will be strongly biased toward the largest pores. Such pores may plug quicker because of the higher degree of polarisation encountered due to the higher flux. (See concentration polarisation.) Also, as pore size increases, the flow rate through the membrane will also increase. The flux for most food solutions is lower than that for water and other pure solvents for a number of reasons. The viscosity of the feed solution may be higher and this further increases throughout the concentration process. Also concentration polarisation and fouling occur. Factors which influence concentration polarisation and fouling are the prevailing flow conditions and the composition of the material being processed. Some of these factors are discussed in more detail in Chapter 4. 3.7 MEMBRANE EQUIPMENT Membrane suppliers now provide a range of membranes, each with different rejection characteristics; for reverse osmosis salt rejection values are quoted, e.g. from 99% rejec- tion of sodium chloride down to 60% rejection of sucrose; and for ultrafiltration, different molecular weight cut-offs in the range 1000 to 500 000. For example, it is implied that a membrane with a molecular weight cut-off of 20000 will reject components with a molecular weight greater than that value. Such figures are provided for guidelines only (see Section 4.2.1). Tight ultrafiltration membranes have a molecular weight cut-off value of around 1000-5000, whereas the more ‘open’ or ‘loose’ membranes have a value in excess of 100 000. Thus molecular weight cut-off value is related to porosity and rejec- tion characteristics; as membranes become more permeable to solutes, their molecular weight cut-off values increase. However, because there are many other factors that affect rejection, molecular weight cut-off should only be regarded as giving a relative guide to its pore size and true rejection behaviour. Experimental determinations should always be made on the system to be validated, at the operating conditions to be used. 76 M.J.Lewis Other desirable features for membranes to ensure commercial success with food com- ponents are listed as Reproducible pore size from batch to batch, offering uniformity in terms of both their permeate rate and their rejection characteristics. High flux rates and sharp rejection characteristics. Compatibility with processing, cleaning and sanitising fluids. Resistance to fouling. An ability to withstand temperatures required for disinfecting and sterilising surfaces, which is an important part of the safety and hygiene considerations. Extra demands placed upon membranes used for food processing include: the ability to withstand hot acid and alkali detergents (low and high pH), temperatures of 90°C for disinfecting or 120°C for sterilising and/or widely used chemical disinfectants, such as sodium hypochlorite, hydrogen peroxide or sodium metabisulphite. The membrane should be designed to allow cleaning both on the feed/concentrate side and the permeate side. See Section 3.8. Membrane processing operations can range in their scale of operation, from laboratory bench-top units, with samples less than 10 ml, through to large commercial scale operations, processing more than 50 m3 h-’. Furthermore, the process can be performed at ambient temperatures, which allows concentration without any thermal damage to the feed components. The process can be batch or continuous; the fluid can be static or in motion, either agitated in a stirred cell or moving across the surface of the membrane. The membrane itself can be configured in a variety of forms. 3.7.1 Membrane configuration The membranes themselves are thin and in most cases require support against the high pressure. The support material itself should be porous. The membrane and its support, together, are normally known as the module. There are a number of criteria that have to be satisfied in the design of a pressure- activated membrane module. It must provide a large surface area in a compact volume; it must be able to support the membrane and the configuration must allow suitable conditions to be established, with respect to turbulence, high wall shear stresses, pressure losses, volumetric flow rates and energy requirements, thereby minimising concentration polarisation. Hygienic considerations are important; there should be no dead spaces and the module should be capable of being in-place-cleaned on both the concentrate and the permeate side. The membranes should be readily accessible, for both cleaning and replacement. It may also be an advantage to be able to collect permeate from individual membranes in the module to be able to assess the performance of each one individually. The two major configurations which have withstood the test of time are the tubular and the flat-plate configurations. The main features of these configurations will be further discussed. Tubular tnembratzes come in a range of diameters. In general tubes offer no dead spaces, do not block easily and are easy to clean. However, as the tube diameter in- creases, they occupy a larger space, have a higher hold-up volume and incur higher Pressure-activated membrane processes 77 pumping costs. The two major types are the holZowJibre, with a fibre diameter of 0.001- 1.2 mm and the wider tube with diameters up to 25 mm, although about 12 mm is a popular size. Some of the configurations are shown in Fig. 3.5 and 3.6. For the hollow fibre system, the membrane wall thickness is about 250pm and the tubes are self-supporting. The number of fibres in a module can be as little as 50 but sometimes greater than 1000. The fibres are attached at each end to a tube sheet, to ensure that the feed is properly distributed to all the tubes. This may give rise to pore plugging at the tube entry point. Prefiltration is recommended to reduce this. The feed is usually pumped through the tubes but it can also be pumped through the shell-side. In general the length of the fibres and modules is not more than 1 m. The geometry of the system ensures that there is a large surface area to volume ratio and that hold-up volumes are low. Pressure drops required are also not excessive because of the short tube lengths. In ultrafiltration applications, the applied pressures are fairly low (usually less than 2 bar). This permits only low inlet pressures, which limits the flow rate that can be achieved as well as restricting its application to low viscosity fluids. However, they are widely used for desalination and in these RO applications are capable of withstanding high pressures. For these applications the outer diameter should be 2-4 times higher than the inner diameter. It is the ratio of the external to internal diameter, rather than the mem- brane wall thickness, which determines the pressure that can be tolerated (Nishimura and Koyama, 1992). Hollow fibre systems usually operate in the streamline flow regime. However, the wall shear rates are high. They may not perform well with viscous feeds and those containing particulate matter. One big advantage is their backflushing capability, which helps improve their cleanability. This is due to the self-supporting nature of the membrane. (b) Fig. 3.5. Plate showing different configurations: (a) and (d) tubular; (b) spiral wound; (c) hollow fibre (with courtesy of PCI). 78 M. J.Lewis (a) Concentrate Drain for backflush Permeate + Entry for backflush Feed --+- r - Drain for backflush (b) Fig. 3.6. Tubular and hollow fibre configurations: (a) tubular, 12.5 mm diameter, 3.6 m long. Module has 18 tubes, total area 2.5 m2 (with courtesy of PCI); (b) hollow fibre, length 1.1 m, diameter 7.6 cm, fibres 0.5 mm or 1.0 mm; total area 1.4 or 2.5 m2. Each module has loo0 or more fibres (with courtesy of Romicon). Hollow fibre systems tend to be expensive, because if one or several fibres burst, the whole cartridge needs to be replaced. For wider tubes, the feed is normally pumped through the tube, which may be up to 25 mm diameter, although a popular size is about 12 mm diameter. There may be up to 20 tubes in one module; tube lengths may be between 1.2 m and 3.6 m (4 and 12 ft) and tubes within the module may be connected in series or parallel. The membrane is cast or inserted into a porous tube which provides support against the applied pressure. Cheryan Pressure-activated membrane processes 79 (1986) describes different systems in more detail. Therefore they are capable of handling more viscous fluids and even materials with small suspended particles, up to one-tenth the tube diameter. They normally operate under turbulent flow conditions with flow velocities greater than 2 m s-'. The corresponding flux rates are high, but pumping costs are also high, in order to generate the high volumetric flow rate required at the operating pressure. They are less susceptible to fouling and relatively easy to clean and disinfect by clean-in-place (CIP) methods. The flat plate module can take the form of a plate and frame-type geometry or a spirally wound geometry. The plate-and-frame system employs membranes stacked together, with appropriate spacers and collection plates for permeate removal, somewhat analogous to plate heat exchangers. The channel height can be between 0.4 and 2.5 mm. Flow may be either streamline or turbulent and the feed may be directed over the plates in a parallel or series configuration. This design permits a large surface area to be incorporated into a compact unit. These systems have been developed for laboratory, pilot plant and large-scale processes, and their performance depends upon the hydrodynamics of the system. One commonly used system commercially is the DDS plate-and-frame system. Here the channel height is between 0.4 and 1 mm. One membrane is fixed on both sides of a permeate collection plate to form a sandwich. Between each sandwich is a spacer plate (see Fig. 3.7(a)). This system usually operates under streamline flow conditions and the energy required is somewhere between that of a tubular and spirally wound system. Membranes are easily replaced and it is easy to isolate any damaged membrane sand- wich. Considerable attention has been devoted to the design of the plate to improve performance. This has been achieved by ensuring a more uniform distribution of fluid over the plate, by increasing the channel width of the longer channels and reducing the ratio of the longest to the shortest channel length. More details are provided by Glover (1985). Also shown in Fig. 3.7(b) is a Pellicon system, used for filtering between 2 and 200 litres. The spiral wound system is now widely used and costs for membranes are quite low. In this case a sandwich is made from two sheet membranes, which enclose a permeate spacer mesh. This is attached at one end to a permeate removal tube and the other three sides of the sandwich are sealed. Next to this is placed a feed spacer mesh and the two together are rolled round the permeate collection tube in the form of a Swiss roll (see Fig. 3.8). The channel height is dictated by the thickness of the feed spacer. Wider channel heights reduce the surface area to volume ratio, but reduce the pressure drop. The typical dimensions of one spiral membrane unit would be about 12 cm in diameter and about 1 m in length. Up to three units may be placed in one housing, with appropriate spacers to prevent telescoping, which may occur in the direction of flow and could damage the sandwich. Another practical aspect which has received attention is excessive by-passing of feed between the periphery of the membrane sandwich and the housing. Reducing this completely by use of a seal at the inlet improves the performance, but causes a dead space, which may be difficult to clean. Meshes or seals which allow limited by-pass provide a reasonable compromise. This configuration is becoming very popular and relatively cheap. Again the flow may be streamline or turbulent. Pressure drop/flow rate relationshim suggest that flow conditions are usuallv turbulent. 80 M.J.Lewis Internal Flow Plates and Spacers 7 'Permeate SPW.?t i-.. Membrane Membrane [-I - * suppoll PI~W Membrane ' Membrane I Flow diagram of Pellicon packet Fig. 3.7. (a) DDS plate arrangement; (b) Pellicon arrangement; with courtesy of DDS and Millipore. Pressure-activated membrane processes 8 1 Anti-Telescoping Device Perforated Tube Feed Solution Membrane Feed Channel Spacer \ \ Permeate Flow Covenng Spirallvound mcmbnnt configuration. Fig. 3.8. Spiral wound configuration (Courtesy of Abcor). Both these flat plate arrangements offer a large surface area to volume ratio and low hold-up volumes, as also does the hollow fibre unit. One consequence of this is that a higher degree of concentration usually results after one pass through the module. How- ever, both tubular and flat plate configurations are used in commercial processing operations. An alternative, much used unit for simple laboratory separations is the stirred cell with agitation facilities. In contrast to systems described earlier, this is a dead-end rather than a flow-through system. It is beyond the scope of this chapter to make comparisons between the different systems and there have been many articles devoted to comparing their performance. However, an overriding consideration is the constraints imposed by the physical proper- ties of the feed material (viscosity and suspended particles) and the extent of concentra- tion required. Other practical considerations such as high flux rates, long processing runs and ease of cleaning are also important. Additional factors that will influence selection are initial capital cost, membrane life, membrane replacement costs, performance with the test solution in question, reliability, power consumption, local availability, and quality of support and after-service. Each system does and will continue to have its devotees. As well as the membrane module, there are pumps, pipeline, valves and fittings, gauges, tanks, heat exchangers, instrumentation and control and perhaps in-place clean- ing facilities. For small installations, the cost of the membrane modules may only be a relatively small component of the total cost of the finished plant, once these other items have been accounted for. This may also apply to some large installations such as water treatment plants, where other separation processes are numerous and the civil engineering costs may also be high. Most membrane plants are built on a modular basis, so it is possible to generate data required for scale-up from a single module, under different operating conditions, in order to determine how flux rate is influenced by pressure, temperature, flow rate and total solids. The effects of fouling and any problems with cleaning can also be established. This type of approach will help to establish the best plant layout, as there are a number of 82 M. J.Lewis options available. Membrane processes can be operated under batch or continuous condi- tions. The simplest system is a batch process. The feed is usually recycled, as sufficient concentration is rarely achieved in one pass. Flux rates are initially high but decrease with time (see Chapter 4). Energy costs are high because the pressure is released each time. Residence times are long. Batch operations are usually restricted to small-scale operations. Batch processing with top-up is used in situations when the entire feed volume will not fit into the feed tank. Batch processing times can be estimated (pre- dicted) if the relationship between flux rate and product concentration is known. In many cases there is a linear relationship between the flux and the log of the Concentration. See Chapter 4. Continuous processes may be single-stage (feed and bleed) or multistage processes, depending upon the processing capacity required. The simplest continuous system is a single-stage process with recycle. Once the retentate has reached its final concentration, a feed and bleed system is operated. Steady state conditions are achieved and product is withdrawn and replaced by fresh feed, at a rate which keeps its composition constant. One drawback is that the concentration in the module is the same as the final product concentration, so the process is operating at its highest total solids content. Therefore, flux rates are low and solids yields may be reduced. The instrumentation and ancillaries for a batch process and a single-stage continuous process are shown in Fig. 3.9. Some of these problems can be alleviated by multistage processes. In multistage processes, the feed may pass once through each stage (single pass), or be recycled within the stage. Single-pass operations are used in situations where a high degree of concentra- tion is achieved in one pass through the stage. Within each stage, the modules may be arranged in parallel or series. The stages are then arranged in series, with a pump or bleed system between the stages. There may be up to six such stages arranged in series in some larger plants. Figure 3.9(c) illustrates a three-stage process. Each stage contains 24 mod- ules, four banks in parallel, each containing six modules in series. One aspect of continuous processing is that the yield is always lower than that for the equivalent batch process. Also the viscosity increases significantly in the end stages and the volume to be processed decreases. Processing plant with an area greater than 200 m2 is now commonplace. Figures 3.10 and 3.1 1 illustrate two commercial installations, based on tubular and plate and frame systems, respectively. 3.8 SAFETY AND HYGIENE CONSIDERATIONS It is important that safety and hygiene are considered at any early stage when developing membrane processes. These revolve round cleaning and disinfecting procedures for the membranes and ancillary equipment, as well as monitoring and controlling the microbial quality of the feed material. For many processes, thermisation or pasteurisation are recommended for feed pretreatment. Microfiltration may also be considered for heat- labile components. From a hygienic viewpoint, the compatibility of the membrane with heat, disinfecting, sterilising and cleaning fluids is critical to the process. There have been considerable advances with this over the last 30 years, since the introduction of jirst- generation membranes based on cellulose acetate. These had limitations; they were not d i - Pressure c - A 84 M. J.Lewis Fig. 3.10. Tubular plant for processing fruit juice (Courtesy of Paterson Candy International). Fig. 3.1 1. Flat plate plant for processing beer (Courtesy of DDS). membranes based on polyamides (low tolerance to chlorine) and particularly polysulphones. These are much more tolerant to acids, alkalis and disinfecting tempera- tures (60-1OO0C), making it possible to clean and disinfect them effectively, to the standards required in dairy processing. They will also withstand 200 ppm chlorine for short periods and 50 ppm chlorine for long-term storage. The cleaning conditions depend upon the type of produce and the membrane used and need to be established experimen- tally for each application. In all cases the usual procedure is to flush out all the product with water before initiating cleaning. For fruit juice clarification, a caustic wash at 55OC, followed by a second caustic wash with 200ppm of hypochlorite, is recommended. Cheryan (1986) reports that hypochlorite helps remove fouling deposits from the pores of Pressure-activated membrane processes 85 the membrane by causing them to swell. Hypochlorite is also a good cold sterilising agent. However, as some organic membranes have a low tolerance to chlorine, care should be taken to abide with the cleaning instructions provided by the membrane supplier. For dairy products, a caustic wash at 50°C to remove fat and protein followed by an acid wash at 50°C to remove minerals is often used. Proteolytic enzymes are useful in cases where heavy protein fouling may be found. Probably the most effective method of assessing a cleaning regime is to restore the water flux to its original value. However, there is evidence that this does not always ensure that the membrane is totally clean. Successful commercial exploitation depends upon ensuring that the membrane can be properly cleaned, the flux restored. Even under these rigorous conditions, membrane lives of well over one year have been reported. After cleaning the membrane needs sanitising or disinfecting. Again a combination of heat treatment (up to 100°C), and cold sanitising fluids based on hypochlorite or hydro- gen peroxide, are available. One problem with chemical cleaning fluids is that they need to be washed out with water. If the membrane is not to be used immediately, a dilute solution of sodium metabisulphite can be used for storage. Relatively little has been reported about the microbiological hazards associated with membrane processing. Yeasts and moulds will cause problems with acidic products, whereas bacteria will require attention with low-acid products. All micro-organisms will be rejected by the membrane and will therefore increase in the concentration, by the concentration factor. There may also be some microbial growth during the process, so the residence time and residence time distribution should be known, as well as’ the operating temperature. If residence times are long, it may be advisable to operate either below 5°C or above 5OoC, to prevent further microbial growth. The feed should be thermised or pasteurised prior to ultrafiltration. Further heat treatment after membrane processing may also be required. Provided the membrane is not damaged, the permeate should be sterile. However, if not properly cleaned, permeate residues will provide nutrients for microbial growth. Membrane units should be designed to allow cleaning and sterilising on both the feed/concentrate and the permeate sides. Although little has been written on the subject, the principles involved in Good Manu- facturing Practice (GMP) (IFST, 1991) and Hazard Analysis Critical Control Points (HACCP) (ICMSF, 1988), can be applied. The philosophy is based on improving safety and quality by prior consideration of all sources of microbial hazards, which could be associated with the raw materials and the methods of processing. The process is analysed in detail and a flow chart is produced. This can be used to identify the potential hazards. Once these have been identified, control criteria can be selected for each hazard, with appropriate monitoring systems, to ensure that the process is under control. Finally it is necessary to document all the procedures and verify that the process is under control by monitoring the microbial quality of the final product. To summarise, the following details should be considered for membrane processes. Raw material quality and treatment: microbial quality, pasteurisation or other heat treatment, filtration, centrifugation. 86 M. J.Lewis Processing conditions: membrane type, operating temperature, pH, residence time and distribution of residence times, all processing parameters should be recorded. Finished product treatment: further heat treatment, storage conditions. Cleaning: temperature, time, detergent concentrations; check to ensure restoration of water flux. Disinfecting/sterilising: times, temperatures, concentrations; check efficacy by taking swabs. Appropriate control criteria for each of these procedures need to be established, together with monitoring systems for ensuring that the process is kept under control. It is also important to verify this. All these procedures should be documented and the end- product should be monitored to ensure that the product meets with the appropriate micro- bial standards. 3.9 REVERSE OSMOSIS APPLICATIONS 3.9.1 Introduction The main applications of reverse osmosis (RO) are for concentrating fluids by removal of water, thereby competing with processes such as vacuum evaporation or freeze- concentration. RO permits the use of lower temperatures even than vacuum evaporation, it avoids a phase change and complete loss of volatiles and it is very competitive from an energy viewpoint. RO uses much higher pressures than other membrane processes, in the range 20-80 bar, and will incur greater energy costs. Suitable high-pressure pumps will be required, which are normally of the positive displacement type, such as piston pumps. These are expensive and contribute a significant component of the capital costs. Areas where evaporation is widely used include the dairy, fruit juice and sugar processing industries. Rejection characteristics for different RO membranes are provided in terms of salt rejection; typically from 80 to 99% rejection of sodium chloride; rejections of other solutes may also be cited, for example calcium chloride and glucose. Products of RO may be subtly different to those produced by evaporation, particularly with respect to low-molecular weight solutes, which might not be completely rejected, and to volatile components, which are not completely lost. (Note that some evaporation plant has aroma recovery facilities.) Reverse osmosis membranes were made for a long time from cellulose acetate. More recently, thin-film composite membranes, based on combinations of polymers, have been introduced, which allow higher temperatures (up to 80°C) and greater extremes of pH (3- 1 1) to be used, thereby facilitating cleaning and disinfection. Note, however, that those based on polyamides have a very low tolerance to chlorine. However, their performance can often be significantly different. For example, Sheu and Wiley (1983) found that the thin film composite membranes were more efficient in retaining flavours than cellulose acetate, during apple juice concentration. There were also differences in salt rejections and organic molecules and these results together with developments in both cellulose acetate and thin-film composite membranes have been covered by Gutman (1987). Pressure-activated membrane processes 87 Therefore the main applications of reverse osmosis are for concentrating liquids, recovering solids and treatment of water. Some of these applications will now be reviewed. 3.9.2 Water treatment Perhaps the most important application is for production of potable water from either brackish water or sea water, by the process of desalination. In this case it is the permeate that is the product of interest. Reverse osmosis is used in many areas world-wide, where there are shortages of fresh water. It has been estimated that by 1987, 25% of desalination capacity world-wide was provided by reverse osmosis (Nishamura and Koyama, 1992). In that year, it was estimated that over 3 000 000 m3 per day of potable water was being produced by reverse osmosis. However, it is still well exceeded by multistage fractional distillation. Potable water should contain less than 500 ppm of dissolved solids. Brackish water (e.g. borehole or river water) typically contains from 1000 up to about 10 000 ppm of dissolved solids, whereas sea water contains upwards of 35 000 ppm dissolved solids; it can be seen that brackish water treatment requires about 95% reduction of solids whereas sea water requires about 99% reduction of solid matter. Sea water is also a more difficult fluid to process and poses more problems than brackish waters, firstly because of the higher osmotic pressure of 40 bar as opposed to about 5 bar for brackish waters and secondly its more serious long-term effects on the membrane performance. Most of the early reverse osmosis units in the late 1960s and early 1970s processed brackish waters, because there were not membranes available to tolerate the high osmotic pressures in- volved with sea water (40 bar). Another interesting feature of both types of plant is that the feed is pretreated by sedimentation, pH adjustment, sand filtration and even microfiltration, in order to reduce fouling to a minimum and ensure that flux reduction is limited by concentration polarisa- tion. Such plant often runs for long periods without the need for intermediate cleaning. Single-pass, multistage designs tend to be favoured for large-scale plants that utilise modules with relatively high water recovery factors, such as spiral and hollow-fibre systems, which are widely used. Brackish water plants usually operate at high water recovery rates (70-95%), so there is a considerable increase in retentate composition toward the end of the process. This gives rise to an increase in the amount of salt which permeates toward the end of the process, so the reduction in solids is not as high as may be expected. Energy costs are about 1-1.5 kWh/m3 permeate. There are many installa- tions processing in excess of 10 000 m3/d. Retentate concentrations are usually kept to below 10 000 ppm, to avoid high osmotic pressures and prevent fouling due to mineral scale. However, more recently the problems involved in treating seawater have been over- come with new membranes that resist compaction at high pressures (up to 70 bar) and show high salt rejections. Some of the earlier installations had a two-stage unit, with some of the permeate being treated in a brackish water type installation. This permeate was then combined with untreated permeate from the first stage to produce an overall product. However, with improvement in membrane performance, it is possible to produce potable water from single-stage processes. Usually seawater is reduced from about 88 M. J.Lewis 35 000 ppm dissolved solids to about 500 ppm. Water recovery rates are much lower than for brackish water treatment (25-30%) and energy requirements are higher (5.6- 7.6 kWh/m3). Reverse osmosis has been used in a wide range of water purification processes, as well as water recovery. Nanofiltration membranes were found to be capable of rejecting a range of synthetic organic compounds from water, although some of the lower molecular weight components evaluated, such as ethylene dibromide, were more permeable. How- ever, a mass balance suggested that some of the higher molecular weight components were adsorbed onto the membrane surface (Duranceau el al., 1992). They have also been investigated for removing pesticides and components responsible for the colour, from ground water, as well as for purifying water for carbonation and soft drinks. For high- grade purity water production, for analytical purposes, it may be treated by double reverse osmosis, as mentioned earlier. Further information on the performance of systems and the economics of the process are given by Gutman (1987) and Nishimura and Koyama (1992). If lower total solids are required, the permeate can be subjected to a second process, known as double reverse osmosis. This has moved on to the production of ultrapure water for the preparation of microelectronic components and uses in medical laboratories and the pharmaceutical industries. The scheme for the production of ultrapure water is given by Gutman (1987) and Nishimura and Koyama (1992). More information is also given in Chapter 1. 3.9.3 Milk processing The potentialities for processing milk by reverse osmosis are not as great as those for ultrafiltration (Grandison and Glover, 1994). It can be used for concentrating full cream milk up to a factor of 2-3 times. Flux decline is similar to that for ultrafiltration, showing a linear relationship when flux is plotted against the log of the concentration factor. Flux rates for skim milk are only marginally higher than those for full-cream milk. Recorded flux rates at the start of the process are up to 40 1 m-* h-'. Factors affecting the flux rate are similar to ultrafiltration. The product concentration attainable is nowhere near as high as that for evaporation, due to increasing osmotic pressure and fouling, due mainly to the increase in calcium phosphate, which precipitates out in the pores of the membrane. Therefore most of the commercial applications have been for increasing the capacity of evaporation plant. Other possible applications that have been investigated and discussed include: the concentration of milk on the farm for reducing transportation costs; for yoghurt produc- tion at a concentration factor of about 1.5, to avoid addition of skim-milk powder; for ice- cream making, also to reduce the use of expensive skim-milk powder; for cheese-making to increase the capacity of the cheese vats, and for recovering rinse water. Whey can also be concentrated, to reduce transportation costs or prior to drying. Flux values for sweet whey are higher than for acid whey, which in turn are higher than for milk, for all systems tested (Glover, 1985). The main reason for differences between acid whey and sweet whey is believed to be the much higher levels of calcium in acid whey, which acts as a foulant. Whey can be concentrated from 6% to 24% solids, at as low as 7°C. A typical plant (PCI technical literature) has a membrane surface area of 327 m2, in three Pressure-activated membrane processes 89 stages, with each stage having 42 modules. Power consumption is 78 kW, the feed rate is 60 m3 h-' and the temperature is 28°C. Under these conditions the membrane lifetime is three years. Suarez et al. (1992) measured the mass-transfer coefficient (k) and the mem- brane concentration c, at different operating pressures and pH values in the range 5.1 to 9.0. They found that both k and c, increase with increasing differential pressure. At high differential pressures, convective flow was high and it was suggested that the process was not mass transfer controlled. Pal and Cheryan (1987) reported some success for using RO concentrated milk (31% TS) for khoa manufacture, with the potential for large savings in energy. However, the average flux was reported as only 8.1 1 m-* h-' at 30°C. Grandison and Glover (1994) reported that for all practical purposes all the components of milk are retained by the membrane and only a small proportion of the smallest ions escape. Rejections of the whole mineral content of milk greater than 99% are reported with rejections of Na+ of 99%, K+ of 98% and C1- of 94%. From a detailed study (Morales et al., 1990), it was found that different membranes and membrane configurations can influence both flux and rejection of components during milk and whey processing. They also found that total solids rejection was independent of temperature and was higher when milk, rather than whey, was processed. In general, all the membranes were capable of rejecting 100% of the true protein. Rejection of non-protein nitrogen, lactose and total BOD was affected by change in the operating conditions, type of feedstock and type of membrane employed, whereas rejection of ash was substantially insensitive to variations in operating conditions and changes in feedstock. Milk concentrate is thus not likely to have the same extent of heat damage as that produced by evaporation. It may also be slightly different in composition, which may affect the texture and stability of products derived from it. For example the storage stability of RO concentrate has been found to be better than that produced by evapora- tion. However, this may not be the ease for all membranes and flow geometries and will depend upon the rejection for the different components. Fouling is a major problem and the main component of the fouling layer is usually found to be protein. Kulozik and Kessler (1988a) considered that there were two resistances to permeation, one due to laminar flow through the deposit and the other due to transport by diffusion through the membrane. They found that the inorganic ions in milk increased the resistance offered by the deposited layer. However, the stability of the deposit and the ease at which it is removed by rinsing is dependent upon the low molecu- lar weight components, particularly calcium (Kulozik and Kessler, 1988b). In sweet- whey processing, fouling resulting from calcium phosphate is a problem at higher concentrations. This can be reduced by prepasteurisation, causing some precipitation of the calcium phosphate or by reduction of pH, using acid or addition of carbon dioxide, where 0.9 mg 1-' has been recommended. One suggestion to reduce the flux decline due to fouling is to control it by use of time-dependent settings of pressure and flow velocity, rather than constant values (Boxtel and von Otten, 1992). Some uses for RO in the dairy, as a water source, are reviewed by the International Dairy Federation (International Dairy Federation, 1988). Nanofiltration has been used for partially reducing calcium and other salts in milk and whey, with typical retention values of 95% for lactose and less than 50% for salts. Guu 90 M. J.Lewis and Zall (1992) have reported that permeate subject to nanofiltration gave improved lactose cry stallisation. NF provides much greater potential for influencing the heat stabil- ity of the milk. Reviews of the use of RO and UF in dairying applications include El-Gazzar and Marth (1991), and Renner and El Salam (1991). 3.9.4 Fruit and vegetable juices Reverse osmosis has found application in the processing of fruit and vegetable juices, sometimes in combination with ultrafiltration and microfiltration. The osmotic pressure of juices is considerably higher than that for milk. There has been a dramatic increase in fruit juice consumption; most juice needs to be concentrated prior to freezing and is then transported frozen. It is advantageous to minimise thermal reactions, such as browning, and to reduce loss of volatiles. From a practical viewpoint, the flux rate and rejection of volatiles is important. RO modules can cope with single-strength clear or cloudy juices and also fruit pulp. RO can be used to produce a final product, as in the case of tomato paste and fruit purees, or to partially concentrate, prior to evaporation. RO is a well-established process for concentrating tomato juice from about 4.5 Brix, to between 8 and 12 Brix. Plant capacity of some commercial units ranges between 25 and 37 m3 h-' and the power consumption is 175 kW. Such plants run at inlet pressures of 50 bar and between 60 and 70°C. The modules are placed in parallel, due to the high viscosity of the product. Tomato juice has been reported to have been concentrated to over 20%, but special care is needed because of the high viscosity (Lafferty, 1992). Other fruit juices which have been successfully concentrated are apple, pear, peach and apricot. Where juices have been clarified, osmotic pressure limits the extent of concentration and up to 25 Brix can be achieved. Unclarified juices may be susceptible to fouling. With purees and pulps, the viscosity may be the limiting factor and these can be concentrated to a maximum of 1.5 times. Gherardi et al. (1989) have used combined UF and RO of pear and peach purees, with the UF permeate being further concentrated by RO, and measured the partitioning of volatile and non-volatile flavour components between the different fractions. Studies by Bowden and Isaacs (1989) have indicated that cloudy pineapple juice can be concentrated by reverse osmosis from about 13 to 25% soluble solids with good quality retention. Losses of soluble components into the perme- ate were found to be very slight. It has also been found useful to recover solids from diluted juice, waste juice and washwater or from depectinised unclarified juice (6-22 Brix). Sheu (1987) reported that the most energy-efficient process for production of apple juice concentrate at 72 degrees Brix, involves processing the clarified juice from ultrafiltration to 21 degrees Brix by reverse osmosis, prior to evaporation. Chou et al. (1991) have studied the loss of flavour compounds during the concentration of apple juice in more detail and concluded that low temperatures (20°C) and high pressures were the most effective for reducing losses. Rejection of flavour compounds was higher using polyamide than polyether-urea thin- film composite membranes. Citrus juices are also concentrated. For oranges, the high hesperin content of the juice results in fouling and rapid flux decline. Performance is improved by pasteurisation and Pressure-activated membrane processes 9 1 filtration to below 0.4 mm. Processing temperatures are between 20 and 40°C and the fouling is reduced by a 10 minute flush with 0.25% caustic soda every six hours. Pectin also contributes to fouling and pectinase treatment has been found to prevent fouling. A process has been described for concentrating orange juice up to 42 Brix, by a combina- tion of UF and RO (Cross, 1989). However, relatively little work has been done on other citrus juices. Braddock et al. (1991) concentrated citrus juice essences containing between 2 and 20% ethanol. There was a substantial reduction in the rejection of ethanol during the concentration of citrus juice essences, which was 90% at 2% alcohol to 40% at 31% alcohol. Acetaldehyde rejection decreased from 65% down to 25% over the same range. Rejection for larger aroma molecules, such as ethyl butyrate, hexanal and terpenes, was generally greater than 85%. The flux rate at 8.3 MPa fell by more than 10 times during concentration from 0.01% to 30% alcohol and flux declined in a linear manner, with the log of the alcohol content. Vegetable juice processing has received some attention, although the market is nowhere near as large as that for fruit juices. Koseoglu et al. (1991b) present data for celery, tomatoes, carrots and cucumbers. The macerated vegetables are pressed and the screened juice is subject to ultrafiltration. The clear permeate can then be concentrated by reverse osmosis and added back to the retentate from ultrafiltration. For all vegetables the flux rates for reverse osmosis were lower than for ultrafiltration and were as follows (1 m-* d-'): carrot, 30.5; celery, 82.5; tomato, 295.7 and cucumber, 432.0. 3.9.5 Other applications Thin-film composite membranes have been assessed for sugar cane and beet juice concentration, up to 80°C and pressures between 40 and 80 bar, Kosikowski (1986). Prefiltration or clarification to remove soil and fibre is essential. Polysulphone mem- branes perform better than cellulose acetate. However, it was pointed out that the potential advantages over evaporation are not so great as there is always plenty of surplus steam available in sugar factories. Instant coffee is a very popular beverage and it is possible to concentrate the coffee extract from about 13% to 36% total solids at 70°C, with little loss of solids. Thin-film composite membranes have been found to give a better retentiofi of aromatics. The concentrate is then evaporated to about 48% solids, prior to drying. Currently, instant tea is also being heavily marketed and reverse osmosis has been investigated for preconcentration. However, much of the research is done within the private sector. From our experience, the tannin components in these drinks are likely to cause fouling prob- lems. Koseoglu et ul. (1991a) have provided a comprehensive review on the application of membrane processes in cereal processing. There is more scope for ultrafiltration, although reverse osmosis is used in combination with ultrafiltration for recovery of protein and other solids from thin stillage materials, which are the remnants after distill- ing the alcohol (Wu, 1987). Again, cellulose acetate membranes were found to be more permeable than polyamide ones. Reverse osmosis is also used for waste recovery and more efficient use of processing water in corn wet-milling processes. More detail is provided by Wu (1988). The process of counter-current reverse osmosis (CCRO) is 92 M. J.Lewis described, which allows concentration to 30% TS, rather than the 20% by normal reverse osmosis. This is achieved by passing a 30% solution on the permeate side of the mem- brane in a counter-current direction to the retentate, thereby reducing the osmotic driving force. A more recent development is the introduction of membranes for concentration of dilute sweetwaters up to 60 Brix, without the use of excessive pressures. Eggs can be concentrated prior to drying. Commercial plant is available for concen- trating egg white, to about 20% solids. In one particular application egg white is concen- trated and dried, after lysozyme has been extracted. It is not crucial to concentrate egg yolk, which contains about 50% solids. Egg processing wastes can also be treated, to reduce BOD, recover solids for animal feed and reutilise water (Roberts, 1989). Poor wine is usually produced from grape juice (must) containing less than 17% sugar. Production of wine from must concentrated slightly by reverse osmosis is improved compared to that produced by adding sugar, although the costs are likely to be higher. Reverse osmosis has been reported to remove some of the compounds responsible for the ‘old’ flavour of wine. It is usually superior to wine produced from evaporated must. Dealcoholisation is an interesting application, using membranes which are permeable to alcohol and water. In a process akin to diafiltration, water is added back to the concentrated product, to replace the water and alcohol removed in the permeate. Such technology has been used for the production of low or reduced alcohol, beers, ciders and wine. It can be applied either as a single process, using a feed and bleed system, or as a two-stage process, where the concentrate from the first stage is rediluted with water and subjected to a second reverse osmosis process. For these applications, cellulose acetate membranes are used rather than the thin-film composites, because their rejection values for ethanol are lower. Gutman (1987) reported that the removal efficiency (rejection) of ethanol was 12% for cellulose acetate membranes and 28% for polyamide membranes. More detailed information on the rejection characteristics and flux data for ethanovwater systems for cellulose acetate, polyamide and other membranes has been collated by Leeper (1986). Ethanol rejections for cellulose acetate ranged between 1.5 and 40%; for polyamides between 32.8% and 60.9% and for other hybrid membranes, as high as 9 1.8%. Further information on the quantitative aspects of diafiltration in terms of how yields and diafiltration volumes are affected by these different rejection values, are provided in Section 4.4. Some methods for producing low-alcohol wines of approxi- mately 5% are described, Chinaud et al. (1991). For beer, quality can be maintained by adding carbon dioxide while it is being concentrated. The use of membranes which reject alcohol could also be used for concentrating beer or fortifying wines. It is also possible to produce low alcohol beverages by use of liquid membranes or using pervaporation (Leeper, 1986). McGregor (1989), has examined the use of a high-flux thin-film composite membrane for concentration of L-phenylalanine from clarified bioreactor harvest media. The rejec- tion coefficient was found to decrease as the retentate concentration increased, and a concentration of 100 g/1 could be obtained. Flux rates were recorded between 17 and 119 1 m-2 h-’. A cascade system could be used to recover almost all the phenylalanine. The author concluded that this study shows the importance of empirical evaluation as the basis of design. Pressure-activated membrane processes 93 In many cases, combinations of processes are used, some of which are discussed in later chapters. One interesting application is the use of formed-in-place membranes (FIP), where the membrane is made by deposition of either inorganic solutes, Thomas ef al. (1992) or organic polymers within the matrix of a porous tube (Spencer and Thomas, 1991). Mechanisms of fouling, a cleaning regime and rejuvenation of FIP membranes are described. SYMBOLS C concentration d pore diameter f concentration factor g acceleration due to gravity h head i degree of ionisation J flux L pore length M molecular weight m ' mass flow rate P pressure AP pressure drop Q volumetric flow rate R rejection R gas constant T absolute temperature V volume V, partial molar volume VRF volume reduction factor W power utilisation Greek symbols Y activity coefficient n osmotic pressure P viscosity e contact angle Subscripts F feed C concentrate P permeate S solute W water m membrane g gel 94 M.J.Lewis REFERENCES Bowden, R. P. and Isaacs, A. R. (1989) Concentration of pineapple juice by reverse osmosis, Food Australia, 41, 850-851. Boxtel, A. J. B. and von Otten, Z. E. H. (1992) In Food Science and Technology Ab- stracts, 5P 67, Braddock, R. J., Sadler, G. D. and Chen, C. S. (1991) Reverse osmosis concentration of aqueous-phase citrus juice essence, J. Food Sci., 56, 1027-1029. Cheryan, M. (1986) Ultrafiltration Handbook, Technomic Publishing Co., Lancaster, USA. Cheryan, M. (1989) Membrane separations: mechanisms and models. In Food Properties and Computer Aided Engineering of Food Processing System, Singh, R. P. and Medina, A. G. (eds.), Kluwer. Cheryan, M. and Nichols, D. J. (1992) Modelling of membrane processes. In Mathemati- cal Modelling of Food Processing Operations, Thorne, S. (ed.), Elsevier Applied Science, London, pp. 45-98. Chinaud, N., Broussous, P. and Ferrari, G. (1991) Application of reverse osmosis to dealcoholisation of wines, Connaissance de la Vigne et du Vin, 25, 245-250. Chou, F., Wiley, R. C. and Schlimme, D. V. (1991) Reverse osmosis and flavour reten- tion in apple juice concentration, J. Food Sci., 56, 484-487. Cross, S. (1989) Membrane concentration of orange juice, Proc. Florida State Horticul- tural SocieQ, 102, 146-152. Duranceau, S. J., Taylor, J. S. and Mulford, L. A. (1992) SOC removal in a membrane softening process, J. Am. Water Works Association, 84, 68-78. El-Gazzar, F. E. and Marth, E. H. (1991) Ultrafiltration and reverse osmosis in dairy technology; a review, J. Food Protection, 54, 801-809. Fane, A. G. and Fell, C. J. D. (1987) A review of fouling control of ultrafiltration, Desalination, 62, 117-136. Gekas, V. (1992) Transport Phenomena of Foods and Biological Materials, CRC Press, Boca Raton. Gheradhu, S., Careri, M., Bolzoni, L., Aldini, R. and Trifiro, A. (1989) Influences of UF and RO on aroma composition of pear and peach purees, Industria Conserve, 64, 199- 206. Glover, F. (1985) Ultrafiltration and reverse osmosis for the dairy industry. Technical bulletin No. 5, NIRD, England. Grandison, A. S. and Glover, F. (1994) Membrane processing of milk. In Modern Dairy Technology, Vol. 1, 2nd edn, Robinson, R. K. (ed.), Elsevier Applied Science, London. pp. 273-3 1 1. Gutman, R. G. (1987) Membrane Filtration, The Technology of Pressure-driven Cross- flow Processes, Adam Hilger, Bristol. Guu, Y. K. and Zall, R. R. (1992) Nanofiltration concentration on the efficiency of lactose crystallisation, J. Food Sci., 57,735-739. ICMSF (1988) Micro-organisms in Foods 4, Application of the Hazard Analysis Critical Control Point (HACCP) System to Ensure Microbiological Safety, Blackwell Scien- tific Publications, Oxford. Pressure-activated membrane processes 95 IFST (1991) Food and Drink-Good Manufacturing Practice: A Guide to its Responsible Management, 3rd ed, IFST, London. International Dairy Federation (1988) The quality, treatment and use of condensate and reverse osmosis permeates. Bulletin No. 232. Jindal, A. R. and Grandison, A. S. (1992) Preparation and composition of chhana whey powders using membrane techniques, J. Sci. Food Agric. 58, 5 11-517. Koseoglu, S. S., Rhee, K. C. and Lusas, E. W. (1991a) Membrane separations and applications in cereal processing, Cereal Food World, 36,376-383. Koseoglu, S. S., Lawhon, J. H. and Lusas, E. W. (1991b) Vegetable juices produced with membrane technology, Food Technol., 45,124, 126-128, 130. Kosikowski, F. V. (1986) Membrane separations in food processing. In Membrane Sepa- rations in Biotechnology, McGregor, W. C. (ed.), Marcel Dekker, New York, pp. 201- 254. Kulozik, U. and Kessler, H. G. (1988a) Permeation rate during reverse osmosis influ- enced by osmotic pressure and deposit formation, J. Food Science, 53, 1377-1383. Kulozik, U. and Kessler, H. G. (1988b) Rinsing behaviour of deposited layers in reverse osmosis, Milchwissenschaft, 43,784-789. Lafferty, I. (1992) Application of membrane systems to the food industry, Paper pre- sented at National Seminar in Food Technology, Mechanisation of Processed Foods, 27-28 October, 1992, Kuala Lumpur, Malaysia, Malaysian Agricultural Research and Development Institute (MARDI). Leeper, S. A. (1986) Membrane separations in the production of alcohol fuels by fermen- tation. In Membrane Separations in Biotechnology, McGregor, W. C. (ed.), Marcel Dekker, New York, pp. 161-200. Lewis, M. J. (1982) Concentration of proteins by ultrafiltration. In Developments in Food Proteins - I, Hudson, B. J. F. (ed.), Applied Science Publishers, Barking. McGregor, W. C. (1989) Concentration of L-phenylalanine with a reverse osmosis mem- brane, J. Biotechnol., 10, 277-283. Medina, B. G. and Garcia, A. I11 (1988) Concentration of orange juice by reverse osmo- sis, J. Food Process Eng., 10,217-230. Morales, A., Amundson, C. H. and Hill, Jr, C. G. (1990) Comparative study of different reverse osmosis membranes for processing dairy fluids, I. Permeate flux and total solids rejection studies, J. Food Proc. Preserv., 14, 39-83. Munari, S., Bottino, A., Capannelli, G. and Moretti, P. (1985) Membrane morphology and transport properties, Desalination, 53, 1 1-23. Nishimura, M. and Koyama, K. (1992) Reverse osmosis. In Membrane Science and Technology, Osada, Y. and Nakagawa, T. (eds.), Marcel Dekker, New York. Pal, D. and Cheryan, M. (1987) Application of reverse osmosis in the manufacture of boa. Process optimisation and product quality. J. Food Sci. and Technol. (India), 24, Renner, E. and El-Salam, M. H. Abd. (1991) Application of Ultrafiltration in the Dairy Roberts, J. J. (1989) Treatment and utilisation of egg processing wastewaters, Disserta- 233-238. Industry, Elsevier Applied Science, London. tion Abstracts International B, 49,4621. 96 M.J.Lewis Sheu, M. J. (1987) Applications of ultrafiltration and reverse osmosis for apple juice clarification and concentration, Dissertation Abstracts International, B, 47, 3 185- 3186. Sheu, M. J. and Wiley, W. C. (1983) Preconcentration of apple juice by reverse osmosis, J. Food Sci., 48,422-429. Spencer, G. and Thomas, R. (1991), Fouling, cleaning, and rejuvenation of formed-in- place membranes, Food Technology, 4598-99. Suarez, E., Martin, F., Alvarez, R. and Coca, J. (1992) Reverse osmosis of whey, J. Membrane Science, 68,301-305. Thomas, R. L., Cordle, C. T., Criswell, L. G., Westfall, P. H. and Barefoot, S. F. (1992) Selective enrichment of proteins using formed-in-place membranes, J. Food Sci., 57, Tsujita, Y. (1992) The physical chemistry of membranes. In Membrane Science and Wu, Y. V. (1987) Recovery of stillage soluble solids from hard and soft wheat by reverse Wu, Y. V. (1988) Reverse osmosis and ultrafiltration of corn-steep liquor solubles, Ce- 1002-1005. Technology, Osada, Y. and Nakagawa, T. (eds.), Marcel Dekker, New York. osmosis and ultrafiltration, Cereal Chem., 64, 260-264. real Chem., 65, 105-109.